1. EGEE is a project funded by the European Union under contract INFSO-RI-508833 Copyright (c) Members of the EGEE Collaboration. 2004. GLUE Schema Sergio Andreozzi JRA1 IT/CZ Cluster sergio.andreozzi@cnaf.infn.it JRA1 All Hands Meeting, 15-17 Nov 2004 www.eu-egee.org

2. JRA1 All Hands Meeting, 15-17 Nov 2004 - 2 Contents Problem Statement GLUE Schema  From the beginning to specification version 1.1  Current Revision Process  What’s new in the current draft for spec version 1.2

3. JRA1 All Hands Meeting, 15-17 Nov 2004 - 3 Problem Statement Resources available in Grid systems must be described in a precise and systematic manner if they are to be able to be discovered for subsequent management or use A shared description allows multiple experts to contribute to the problem and serves as a communication mean between different knowledge domains INFORMATION MODEL Abstraction of real world into constructs that can be represented in computer systems (e.g., objects, properties, behavior, and relationships)  Not tied to any particular implementation  Used to exchange information among different domains

4. JRA1 All Hands Meeting, 15-17 Nov 2004 - 4 Problem Statement Main Use Cases:  Discovery for brokering and access: “what are the Computing Elements available to the VO CMS and that offer the SL3 operating system with installed the CMKIN software package?” “what are the Storage Elements that offer 20 gigabytes of disk space for the VO ATLAS?”  Discovery for monitoring “how many CPUs the site XYZ is offering to the EGEE Grid?” “what is the success rate of job submitted per site?”

5. JRA1 All Hands Meeting, 15-17 Nov 2004 - 5 GLUE Schema approach to the information modeling of Grid resources started in April 2002 by the DataTAG and iVDGL projects Contributions from DataGrid, Globus, PPDG, GryPhyn Globus Schema (LDAP) DataGrid Schema (LDAP) GLUE Schema (UML) GLUE Schema (Relational) R-GMA GLUE Schema (XML) GT MDS 4 GLUE Schema (LDAP) GT MDS 2

6. JRA1 All Hands Meeting, 15-17 Nov 2004 - 6 GLUE Schema – current spec 31 March 2003: GLUE Schema version 1.1  Computing Resources Computing Element Cluster SubCluster Host  Storage Resources Storage Element Storage Space Data Access Protocol Storage Library  Computing/Storage Relationship CESEBind [ 1 ]

7. JRA1 All Hands Meeting, 15-17 Nov 2004 - 7 GLUE Schema Current Revision Process

8. JRA1 All Hands Meeting, 15-17 Nov 2004 - 8 GLUE Schema – next steps There is a number of issues to be addressed:  Simplification  “bug fixing”  better documentation  new use cases  Extensions: Per-VO view of computing resources (e.g., Estimated Traversal Time, Free Job Slots) (Jeff Templon) Grid3 (some already in the new draft) GridICE (Grid monitoring) INFN (Monitoring the Connectivity of a Grid) JRA1 Advance Reservation (being defined with T. Ferrari and E. Ronchieri) [ 5 ] [ 4 ]

9. JRA1 All Hands Meeting, 15-17 Nov 2004 - 9 GLUE Schema – current revision process Communication: Mailing list: glue-schema@hicb.orgglue-schema@hicb.org Tracker: http://infnforge.cnaf.infn.it/tracker/?atid=118&group_id=9&func=browse http://infnforge.cnaf.infn.it/tracker/?atid=118&group_id=9&func=browse Phone Calls or face-to-face meeting are not scheduled since one year Revision Process: Items for schema revision can be added in the tracker by anyone People can post their comments in the tracker  Each tracker update is forwarded to the mailing list When an agreement is reached, the change is inserted in the next schema revision

10. JRA1 All Hands Meeting, 15-17 Nov 2004 - 10 GLUE Schema – current revision process Even though the active participation in this activity is still a problem, there is a periodical expression of interest in this work and its evolution In order to go ahead, each involved project should refresh who are the representatives that should actively participate and make decision on behalf of the project Area of interests vs. resouce cateories  we should propose names for EGEE ComputingStorageNetwork Brokering  Data Management  ? Advance Reservation  Monitoring 

11. JRA1 All Hands Meeting, 15-17 Nov 2004 - 11 GLUE Schema – next steps Possible approach:  Minor revision taking into account only bug fixing and extensions that maintain backwards compatibility (e.g., GLUE Schema 1.2 draft)  Major revision including refactoring for simplification (e.g., GLUE Schema 2.0 to be written)  LCG2 should be interested in the minor revision  gLite should consider the major one [ 3 ]

12. JRA1 All Hands Meeting, 15-17 Nov 2004 - 12 GLUE Schema – The importance of Use Cases Use Cases are important  They are two-way: Check if we are able to represent the reality Check if what is in the model can be measured So far, we lack a Use Case document People participating in the deployment should help in collecting a number of significant use cases Each schema revision should be checked against them

13. JRA1 All Hands Meeting, 15-17 Nov 2004 - 13 GLUE Schema – The importance of Use Cases an example Site A has 6 worker nodes (3 fresh new and fast, 3 old and slow) The farm is configured as follows:  a high-end queue to the 3 fast WN’s  a slow queue to the 3 slow WN’s  a background queue to the 6 WN’s (lower priority) Queue Host - Fast Host - Slow Cluster

14. JRA1 All Hands Meeting, 15-17 Nov 2004 - 14 GLUE Schema – The importance of Use Cases an example Representation in the GLUE Schema v. 1.1 Queue -> Computing Element (CE_highend, CE_slow, CE_background) Characteristics of new/fast WN’s -> Subcluster A Characteristics of old/slow WN’s -> Subcluster B A+B=Cluster Problem: there is no explicit relationship between CE and SubCluster; CE are associated to Cluster, hence to all underlying SubClusters Queue Host - Fast Host - Slow Cluster X X

15. JRA1 All Hands Meeting, 15-17 Nov 2004 - 15 GLUE Schema – The importance of Use Cases an example Possible Fix in GLUE Schema v. 1.2 (Additive approach) Add a new relationship: CE (*) ---can submit to--- (*) SubCluster

16. JRA1 All Hands Meeting, 15-17 Nov 2004 - 16 GLUE Schema – The importance of Use Cases Possible Major Refactoring in GLUE Schema v.2.0  CE is a site cluster (EGEE Architecture document)  Queues are used to differentiate the service  The service offers access and management of available execution environments ServiceClass:  HighEnd, Slow, Background ExecutionEnvironment:  charact node A, charact node B

17. JRA1 All Hands Meeting, 15-17 Nov 2004 - 17 GLUE Schema Version 1.2 Draft proposal

18. JRA1 All Hands Meeting, 15-17 Nov 2004 - 18 GLUE Schema 1.2 – Core New/modified Unmodified Deprecated

19. JRA1 All Hands Meeting, 15-17 Nov 2004 - 19 GLUE Schema 1.2 – Computing New/modified Unmodified Deprecated

20. JRA1 All Hands Meeting, 15-17 Nov 2004 - 20 GLUE Schema 1.2 – Cluster/SubCluster/Host New/modified Unmodified Deprecated

21. JRA1 All Hands Meeting, 15-17 Nov 2004 - 21 GLUE Schema 1.2 – Storage New/modified Unmodified Deprecated

22. JRA1 All Hands Meeting, 15-17 Nov 2004 - 22 GLUE Schema 1.2 – CE/SE relationship New/modified Unmodified Deprecated

23. JRA1 All Hands Meeting, 15-17 Nov 2004 - 23 Bibliography [1] GLUE Schema Resources: http://www.cnaf.infn.it/~sergio/glue [2] GLUE Schema – open issues: http://infnforge.cnaf.infn.it/docman/view.php/9/65/GLUEInfoModel_V_1_2_draft_1.pdf [3] GLUE Schema Version 1.2 – First Draft http://infnforge.cnaf.infn.it/tracker/?atid=118&group_id=9&func=browse [4] Grid3 metrics http://grid.uchicago.edu/metrics/metrics-table.html http://griddev.uchicago.edu/download/grid3/doc.pkg/monitoring-metrics/Grid3-metrics.doc [5] Monitoring the Connectivity of a Grid. S. Andreozzi, A.Ciuffoletti, A. Ghiselli, C. Vistoli. In Proc. of the 2nd International Workshop on Middleware for Grid Computing (MGC 2004) in conjunction with the 5th ACM/IFIP/USENIX International Middleware Conference, Toronto, Canada, October 2004.